Integrating remote sensing and topographic modeling to identify rainfall-induced water accumulation zones in man-made environment of Kassandra, Greece

Rapid and reliable methodologies for identifying areas susceptible to rainfall-induced water accumulation are necessitated by frequent flood events in downstream regions. These techniques are especially vital where such zones intersect with critical infrastructure. For that purpose, remote sensing, Geographic Information Systems (GIS), and topographical analysis are usually integrated within a methodological framework to detect flood-prone areas. This research focuses on the man-made environment of Kassandra, Greece, while emphasizing road infrastructure. Sentinel-2A satellite imagery was utilized by Object-Based Image Analysis (OBIA) and Support Vector Machine (SVM) algorithms for land cover classification. An overall land cover classification accuracy of 84% was obtained. The Topographic Wetness Index (TWI) was calculated from ALOS digital elevation model data to identify areas with high water accumulation potential, and values exceeding 9 were classified as critical. A negative correlation (-0.24) between the Normalized Difference Tillage Index (NDTI) and TWI was revealed by statistical analysis. This result confirmed that areas with elevated water accumulation potential correspond to man-made surfaces. The results showed that approximately 19.5% of the total man-made environment and 21% of the road network are located within zones susceptible to water accumulation. Spatial analysis identified critical intersections between road segments and stream networks, which provides actionable intelligence for flood risk mitigation. However, digital elevation model resolution limits the accuracy of the methodology. Thus, future risk assessment would benefit from the integration of additional hydrological variables.